Re: Requirements for (level >=3) tests

Le Jeu 16 février 2012 13:38, Aryeh Gregor a écrit :
> 2012/2/16 "Gérard Talbot" <css21testsuite@gtalbot.org>:
>> I disagree with you.
>
> Okay.
>
>> XHTML tests are not served as text/html in the test suite.
>
> Correct.  This just means that you're using a different code path from
> virtually every web page, which is undesirable in a test suite.  Tests
> should be text/html because web pages are all text/html, for practical
> purposes.

The XHTML format insures that they can be later built into well-formed
HTML4. There are reasons why tests should be created in XHTML served as
application/xhtml+xml to begin with.

Personnally, I never used and never use XHTML served as
application/xhtml+xml .


>
>> XHTML insures well-formedness.
>
> This begs the question.  Why do we want well-formedness?  HTML parsing
> has been well-defined for several years now.


"Tag soup" is definitely not what I propose, recommend.


>
>> What's so difficult in using a HTML document template anyway? Most
>> text
>> editors can be customized to start with a HTML document template.
>
> It's harder to read.  As in, several times as long, with most of the
> markup being boilerplate that serves no purpose.  It's processed
> exactly the same by all browsers, and the relevant standards guarantee
> that.  The HTML5 version of the same file is two lines long and
> contains nothing except the actual style information relevant to the
> test, except the doctype.


So the author name in that reftest is not important, useful? Who are web
authors supposed to contact if there is a problem with a reftest then?


>
>> The whole thing sends a wrong message to web authors out there that
>> web
>> standards are not important to follow, web standards bring no
>> benefits,
>> are useless anyway, pointless, etc.. Web authors could now look at the
>> tests and say: "Look, even W3C in the test suites does not follow the
>> web standards it creates."
>
> HTML5 is a W3C-hosted specification, and the test file I gave is valid
> HTML5 (except that it's missing a <title>, okay).


It is invalid HTML5. It's not just missing a <title>. It is invalid
HTML5 because it is missing the <title>. And the W3C validator will
report warnings about undeclared character encoding too.
According to W3C quality assurance tips for web authors, <title> is the
most important element of a quality Web page
http://www.w3.org/QA/Tips/good-titles
and now we should just remove it from testpages and reftests of W3C test
suites?


> XHTML 1.0 is a
> specification that's more than a decade old, is exceptionally vague in
> specifying everything except syntax, and isn't actually used on a
> nontrivial number of real-world websites (given that pages served as
> text/html are actually processed by browsers as HTML5 regardless of
> doctype).  Using it in preference to HTML5 sends the message that the
> W3C cares more about theoretical purity than about actual browser
> interoperability.  Which, unfortunately, seems to be true.

A few min. ago, I wrote about &gt; and child selector. Now, if everyone
had followed the CSS format guidelines, then we would never have
encountered such issue. So, in this case (using <style
type="text/css"><![CDATA[  ...  ]]></style> ), it was not and it is not
about theoretical purity.

Web standards exist so that problems, issues, difficulties, etc. can be
avoided and interoperability can prevail.

Gérard
-- 
Contributions to the CSS 2.1 test suite:
http://www.gtalbot.org/BrowserBugsSection/css21testsuite/

CSS 2.1 Test suite RC6, March 23rd 2011:
http://test.csswg.org/suites/css2.1/20110323/html4/toc.html

CSS 2.1 test suite harness:
http://test.csswg.org/harness/

Contributing to to CSS 2.1 test suite:
http://www.gtalbot.org/BrowserBugsSection/css21testsuite/web-authors-contributions-css21-testsuite.html

Received on Thursday, 16 February 2012 19:05:19 UTC